Live Demo

MailPilot: Live Demo

A live AI agent that reads email, searches a knowledge base, and responds — fully automated.

demo@lab5.ca
Send an email →

A live AI agent responds to real emails. No signup required.

How the Demo Works

  1. 1

    Send a product question

    Email demo@lab5.ca with a question about the product specs in the knowledge base. Questions outside the knowledge base will receive a reply explaining how the demo works.

  2. 2

    The agent searches the knowledge base

    An AI agent searches product documentation stored in Google Drive, using a RAG pipeline to find relevant specifications.

  3. 3

    You get an accurate reply

    Within 60 seconds, you receive a professional response grounded in real data. The entire flow — from Gmail notification to reply — is fully automated.

Product Specifications

These questions are based on real product specification sheets. Review the source documents to verify the agent's accuracy.

Google Drive

Try These Questions

Questions the agent can answer

These questions match products in the knowledge base. The agent will reply with accurate, sourced answers.

  • "What are the dimensions and weight of the TW-18.0K-1240 reverse osmosis system?"
  • "Which SF-100S softener would you recommend for a hospital needing at least 200 GPM continuous flow?"
  • "Which UV-COM model supports the highest flow rate, and what certifications does it have?"
  • "What happens if feed water TDS exceeds 5,000 ppm on the RO-200 series?"
  • "Compare the WS48-1200 and WS54-1500 softener models in capacity and flow rate."
  • "I need to treat hard water at 40 GPG for a commercial building. Do I need a softener before the RO system, or can the UV-COM handle it?"
×

Questions the agent will decline

These products are not in the knowledge base. The agent will tell you it doesn't have that information rather than guess.

  • "What is the price of the Pentair Evoqua chlorine dioxide generator?"
  • "Can you recommend a Grundfos booster pump for a 20-story building?"

Under the Hood

For technical reviewers — here's how MailPilot is built and why I made these choices.

Frontend

Astro SSG, Tailwind CSS, edge deployment. Static generation for zero cold-start latency, deployed globally.

Backend

Python, serverless functions, event-driven messaging. Architecture triggered by email push notifications. Serverless for scale-to-zero costs.

AI Layer

LLM integration, knowledge base search. RAG pipeline finds relevant documentation. Structured prompts guide multi-turn email conversations.

Imagine this for your business

MailPilot shows my approach: find a real business problem, automate it with AI, and deploy to production. What repetitive process is slowing your company down?